Is it necessary to filter based on missing rate?
I noticed that you mentioned in #29 that "SMC++ assumes that the data has not been filtered on the basis of mutation frequency". At the same time, #266 states that "SMC++ combines the computational efficiency of the SFS and the advantage of utilizing LD information in coalescent HMMs."